LSIDs in Taverna

1
LSIDs in Taverna

• Daniele Turi
• University of Manchester
• RDF, Ontologies and Metadata, Edinburgh, 7-9/6/06

2
Outline

• Taverna Workbench
• workflows of biological services
• LSIDs used to identify
• data, workflows, workflow runs
• LSIDs and Named Graphs
• LSID Resolution
• Security (under development)
• LSID granularity

3
myGrid

• eScience project
• biological workflows
• compose web services
• execute
• discover
• audit/provenance

4
myGrid

• eScience project
• biological workflows
• compose web services
• execute
• discover
• audit/provenance

Taverna
5
myGrid

• eScience project
• biological workflows
• compose web services
• execute
• discover
• audit/provenance

Taverna
Provenance
Service Annotation/ Discovery
6
Taverna Workbench

• Large user community in biology
• about 1,000 downloads per month
• one release each 6 weeks
• Collect and browse provenance
• new feature (released 2 days ago!)

7
Taverna Workbench
8
Provenance as RDF

• RDF generated automatically
• audit trail
• RDF is typed (semantics!)
• 1 RDF graph for each workflow run
• named graph

9
Workflow Run
urnlsidworkflow6
urnlsidorgHY7
runs
belongsTo
urnlsid..wfRunHU77I8
launchedBy
urnlsidperson4
hasInput
hasInput
urnlsiddataItemK84P
urnlsiddataItem51HJ3
10
Typed Workflow Run
launchedBy
Provenance Ontology
hasInput
Experimenter
Organization
DataObject
WorkflowRun
Workflow
belongsTo
runs
urnlsidworkflow6
urnlsidorgHY7
runs
belongsTo
urnlsid..wfRunHU77I8
launchedBy
urnlsidperson4
hasInput
hasInput
urnlsiddataItemK84P
urnlsiddataItem51HJ3
11
(No Transcript)
12
(No Transcript)
13
LSIDs

• LSIDs used to identify
• data, workflows, workflow runs
• internal
• external LSIDs not used (call by value)
• Taverna 2 (call by reference) near future
• data and workflows (and people and
  organizations!)
• Workflow runs LSIDs are names of graphs

14
Storage

• Named RDF graphs
• retrieve whole workflow runs
• implementation in
• Sesame2 native store
• scalable
• alpha release (bugs)
• NG4J (Jena MySQL)
• scalability issues
• Future implementations Oracle and Boca

15
LSID Resolution

• Implemented but not deployed ?
• obstacle single user v enterprise
• virtual organisation
• Resolution returns
• only data for workflows and data
• only metadata for workflow runs
• Data v Metadata
• why data immutable and metadata mutable?

16
Security

• LSID granularity very good
• Policies (in XACML) easily expressed in terms of
  LSIDs
• LSID spec does not mention https and credentials
• IBM Java Toolkit supports credentials

17
Security Policy

• Scenario
• supervisors can access all workflow runs in the
  organization
• students can access only their own workflow runs
• blacklisted users cannot access anything
• See policySet.xml on myGrid wiki

18
(No Transcript)
19
Conclusions

• LSIDs ?
• Named Graphs ?
• persistence ?
• Ontologically typed RDF ?
• Mutable v immutable identified with metadata v
  data ?
• Credentials not part of LSID spec ?
• LSID granularity for security ?